Splitting frequency of the (2 + 1)-dimensional Duffin-Kemmer-Petiau oscillator in an

external magnetic field

Ignacio S. Gomeza,, Esdras S. Santosa , Olavo Ablaa

a Instituto de Fı́sica, Universidade Federal da Bahia, Rua Barao de Jeremoabo, 40170-115 Salvador-BA, Brazil

Abstract
We revisit the (2+1)-dimensional DKP oscillator in an external magnetic field by means of 4×4 and 6×6 representations of the DKP
field, thus obtaining several cases studied in the literature. We found an splitting in the frequency of the DKP oscillator according
to the spin projection that arises as an interplay between the oscillator, the external field and the spin, from which the energies and
arXiv:2005.01228v2 [math-ph] 2 Jul 2020

the eigenfunctions are expressed in a unified way. For certain critical values of the magnetic field the oscillation in the components
of spin projections -1 and 1 is cancelled. We study the thermodynamics of the canonical ensemble of the vectorial sector, where
a phase transition is reported when the cancellation of the oscillation occurs. Thermodynamic potentials converge rapidly to their
asymptotical expressions in the high temperature limit with the partition function symmetric under the reversion of the magnetic
field.
Keywords: DKP oscillator, splitting frequency, relativistic wave equations, vectorial sector thermodynamics

Introduction calculated the eigensolutions of massive spin-0 and spin-1 par-

ticles both in commutative and non-commutative phase-space.
The Duffin-Kemmer-Petiau (DKP) equation is a relativis- In this letter we revisit this problem in commutative space in
tic first-order wave equation which describes scalar and vector order to show some interesting properties in the spin one sector,
fields through a unified formalism [1–4]. Using its represen- which can be found with suitable conditions for components of
tations [5, 6] or projectors [7, 8] as well as a rich variety of the DKP field. The work is organized as follows. We begin by
couplings [9] for the spin zero and spin one sectors, this the- considering a scalar 4×4 and a 6×6 representations for the DKP
ory has been applied to the study of many different problems: field in the presence of an external magnetic field, that allows
the scattering of K + nucleus [10], quantum chromodynamics to study several cases of interest in the literature. Then, we ana-
[11], studies on the S-matrix [12], causality of the DKP the- lyze the scalar and vectorial cases as well as some special ones,
ory [13], minimal and non-minimal coupling in a general rep- thus obtaining the simplified DKPO and the two-dimensional
resentation of DKP matrices [14], Bose-Einstein condensation DKPO along with their energies, eigenfunctions and degenera-
[15, 16], breaking of Lorentz symmetry [17], curved space- cies. In the spin one sector of the DKP representation we found
time [18], Aharonov-Bohm potential [19], studies of the phase a relationship (splitting) between the frequency of the DKP os-
in Aharonov-Casher effect [20], Galilean five-dimensional for- cillator and the projection spin of the particles, whose main ef-
malism [16, 21–23] and several others applications [24–38]. fects are to flip the vectorial components s1,2 = ±1 when the
Particularly, in [39] the authors have shown that the (2+1)- direction of the magnetic field is reversed and to cancel the os-
dimensional Dirac oscillator with frequency ω in an external cillation in the s1,2 component for some values of the magnetic
magnetic field taken along the z-direction and represented by field. Next, we study the statistical properties of the canoni-
the vector potential A = B2 (−y, x), can be mapped onto the cal ensemble of the vectorial sector and we derive its thermo-
Dirac oscillator without magnetic field but with reduced angu- dynamics, exhibiting phase transitions for some values of the
lar frequency given by ω − ω e, where ωe = 2mc
eB
and e and m are magnetic field, as a consequence of the splitting. Finally, we
the charge and the mass of the electron. This result has been summarize the results and draw the conclusions.
used, for example, to study atomic transitions in a radiation field
[39] as well as for its corrections through the generalized uncer-
tainty principle [40]. Following the same approach, the (2+1)-
DKP equation and some representations
dimensional DKP oscillator (DKPO) has been studied under an
external magnetic field [31, 33, 34, 38], where the authors have
The DKP equation is written by the form [1]-[3] (see the
∗ Corresponding
beautiful review by Krajcik and Nieto [4] and the references
author
therein)
Email addresses: [email protected] (Ignacio S.
Gomez), [email protected] (Esdras S. Santos),
i~βµ ∂µ − mc Ψ = 0,
 
[email protected] (Olavo Abla) (1)
Preprint submitted to Elsevier July 3, 2020
where Ψ is a DKP field of mass m and β-matrices satisfy the Scalar DKPO in a magnetic field
DKP algebra
For the scalar sector, using a four components ΨT =
qB
βµ βλ βν + βν βλ βµ = βµ ηλν + βν ηλµ . (2) (Ψ0 , Ψ1 , Ψ2 , Ψ3 , ) and ω
e = 2mc , the coupled equation (6) pro-
vides the following four equations
As predicted by [5] and showed in [6], the DKP equation in
mc2 Ψ0 = EΨ3 (7)
(2 + 1)-dimensional space-time has two representations: a 4 × 4
for the scalar sector and a 6 × 6 for the vector sector, with the mc2 Ψ1 = −c (p x + me
ωy − imωx) Ψ3 (8)
 
Lorentz metric gµν given by g00 = −g11 = −g22 = 1. The DKP mc Ψ2 = −c py − me
2
ω x − imωy Ψ3 (9)
representation for the scalar sector is given by matrices mc Ψ3 = EΨ0 + c (p x + me
2
ωy + imωx) Ψ1
 
 ·

· · 1 
 
 · · · · 
 +c py − me
ω x + imωy Ψ2 . (10)
 · · · ·  1 

· · · 1 

β = 
0  , β =  , Performing the combination of these equations we have
 · · · ·   · · · ·   
1 · · · · −1 · · E 2 − m2 c4 Ψ3 = c2 (p2 + m2 ω20 r2 − 2m~ω − 2me ωLz )Ψ3 , (11)
 · · · ·  √
 
 · · · · 
 where ω0 = ω2 + ω e2 , r2 = x2 + y2 and Lz = xpy − yp x is
β = 
2 , (3) the z-component of the angular momentum. This result is in
 · · · 1 
· · −1 · agreement with [33], for a null non-commutative factor. Thus,
following the standard approach, the energy spectrum is given
and the DKP representation for the vector sector is given by the by
6 × 6 matrices r
2~(|l|eω + ω) ~ω0
 ·

· · −1 · · 
 En,l = ±mc 2
1− + (4n + 2(|l| + 1)), (12)
 mc2 mc2
 · · · · −1 · 

 · · · · · ·  where n = 0, 1, 2, 3, ... is the quantum number associated to
β0 =  , the energy and l = 0, ±1, ±2, ±3... is the angular momentum
 −1 · · · · · 
 · −1 · · · · 
 quantum number. The eigenfunction Ψ3 is calculated using the

· · · · · ·
 polar coordinates (r, θ), we have
 mω |l|/2 mω0 2
 mω 
0 2 0 2
(Ψ3 )n,l = r eilθ e− 2~ r Ln|l| r , (13)

· · · · · 1
 ~ ~
 
 · · · · · ·  where Ln stands for the nth Laguerre polynomial and the non-
taking E =  + mc2 with
 
· · · · 1 · relativistic limit of (12) is calculated
β =   ,
1
  
 · · · · · ·   << mc2 giving E 2 − m2 c4 /2mc2  ε. The others compo-
· · −1 · · · nents of the DKP eigenfunction, Ψ, can be easily calculated us-
 
 
−1 · · · · · ing the expressions (7)-(9). These results show a perfect equiv-
alence between the representations formalism [6] and the one
employed on the magnetic DKP oscillator from the past years
· · · · · · 
 
  [31, 33, 34, 38], considering the scalar sector and the commuta-

 · · · · · 1  tive phase-space. Next, we want to show the advantage of this
· · · −1 · · 
β =  .

2
(4) formalism, with no particular choice to describe the system like
 · · 1 · · ·  was made before, where the expected results are the spin effects
· · · · · · 
 
  and a curious “splitting” on the frequencies.
· −1 · · · ·

The DKPO in the presence of an external magnetic field A = Vector DKPO in a magnetic field: splitting in the frequency
B
2
(−y, x) is performed by the coupling [31, 33, 34, 38]
For the vectorial sector, using the equation (B3) of [14] and
q
p → p − A − imωη0 r, (5) taking R0 Ψ = 0 (thus making the scalar component equal to
c zero), the equation (49) of [14] is simplified by cancelling the
where ω is the angular frequency of the oscillation, q is the term proportional to 1/m2 . In this way it is obtained a simpli-
 2 fied equation for the (3 + 1)-dimensional vector DKP oscillator
charge of the boson and η0 = 2 β0 − 1. The coupled DKP (SDKPO), which is composed by the usual three-dimensional
equation is written by oscillator added to a spin-orbit coupling. Following this rea-
  q   soning, in the (2 + 1)-dimensional case we choose conveniently
−β0 E + cβ · p − A − imωη0 r + mc2 Ψ = 0, (6) the six components of the DKP field ΨT = (a, b, d, 0) where
c
  a = (a1 , a2 ) and d = (d1 , d2 ). The null component was cho-
where we have used −i~∂µ = −pµ = Ec , p . sen in order to obtain the (2 + 1)-dimensional SDKPO. Using
2
Ψ, from the representation (4) as well as the coupling (6) in the
(1), we obtain the six equations e
oscillator and field angular frequencies ω, ω

mc2 a1 = −Ed1 (14) e s1 = 1

ω1 = ω + ω
mc a2 = −Ed2
2
(15) √
splitting e 2 s0 = 0
ω0 = ω 2 + ω
mc b = −c (p x + me
2
ωy − imωx) d2
e s2 = −1
ω2 = ω − ω
 
+c py − me ω x − imωy d1 (16)
 
−mc2 d1 = Ea1 + c py − meω x + imωy b (17)
Figure 1: Schematic representation of the frequency splitting of the DKPO
−mc2 d2 = Ea1 − c (p x + ime
ωy + imωx) b (18) (Eq. (31)) in the presence of an external magnetic field, according to the spin si
0 = c (p x + meωy − imωx) a1 − with i = 0, 1, 2.
 
c py − me ω x − imωy a2 . (19)
By substituting Eqns. (14) and (15) into the others, we have the and

equations for the components b, d1 and d2 , given by
(E 2 − m2 c4 )ϕ2 = c2 p2 + m2 (ω − ω
e)2 r2
(E 2 − m2 c4 )b = c2 (p x + me
ωy − imωx)(p x + ime
ωy + imωx) 
+c2 (py − me
ω x − imωy)(py − me
ω x + imωy)]b (20) −2m~(ω − ω e) + 2m(ω − ω e)Lz ϕ2 . (30)
(E 2 − m2 c4 )d1 = −c2 (py − me
ω x + imωy) ×
Equations (29) and (30) can be rewritten in the more compact
[(p x + me
ωy − imωx)d2 − (py − me
ω x − imωy)d1 ] (21) form
(E 2 − m2 c4 )d2 = c2 (p x + me
ωy + imωx) ×    
E 2 − m2 c4 ϕi = c2 p2 + m2 ω2i r2 − 2m~ωi − 2mωi Lz S z ϕi , (31)
[(p x + me
ωy − imωx)d2 − (py − me
ω x − imωy)d1 ] (22)
0 = (p x + me
ωy − imωx)d1 + (py − me
ω x − imωy)d2 . (23) where i = 1, 2 with ω1 = ω + ω e, ω2 = ω − ω e and S z ϕi = ~si ϕ
with s1 = −s2 = 1. The last term 2ωLz in the right side of (31)
Replacing (21) and (22) in (20) it follows that plays the role of a spin-orbit term. Following this reasoning, we
can interpret the functions ϕi as components of a vector (ϕ1 , ϕ2 )
  h i
E 2 − m2 c4 b = c2 p2 + m2 ω20 r2 + 2m~ω − 2me ωLz b, (24)
√ whose spin projections along the z-direction are +1 and −1, re-
where ω0 = ω2 + ω e2 . The expression of above show us that spectively. Thus, in the presence of an external magnetic field,
the component b, like (11), behaves as a scalar DKPO in a mag- the spin one DKPO presents a scalar component, b, and two
netic field. The energy spectrum for (24) is obtained by the ex- vector components, ϕ1 and ϕ2 , which oscillate with the angular
change ω → −ω in (12), while its eigenfunction is equal to (13). frequencies ω0 , ω1 and ω2 , respectively. The expression (31)
In other hand, if we apply the operator c2 (p x + me ωy + imωx) has the form of the equation for the (2+1)-dimensional SDKPO
to (23) and add the result to (21), and then we apply the opera- whose energy spectrum and eigensolutions are
tor c2 py − me ω x + imωy to (23) and add the result to (22), we r
have ~ωi
En,l = ±mc
(i) 2
1+ (4n + 2|l| (1 − si ) + 2si ) i = 1, 2 (32)
     mc2
E 2 − m2 c4 d1 = c2 p2 + m2 ω2 + ω e2 r2 − 2m~ω
 where n = 0, 1, 2, 3, ..., l = 0, ±1, ±2, ±3... with En,l
(1) (2)
and En,l
the energies associated to the angular frequencies ω1 and ω2 ,
 
−2me ωLz d1 + ic2 2m~e ω + 2mωLz − 2m2 ωe ωr2 d2 (25)
respectively. The eigenfunctions ϕi are expressed by
and  mω |l|/2
ilθ − 2~i r2 |l| mωi 2
mω
 
i 2
ϕ(i) = ,

r e e L r (33)
   
E 2 − m2 c4 d2 = c2 p2 + m2 ω2 + ω
e2 r2 − 2m~ω n,l ~ n
~

−2me

ωLz d2 − ic2 2m~eω + 2mωLz − 2m2 ωe

ωr2 d1 . (26) whose non-relativistic limit is calculated by the same way as
pointed out previously. The others components of the spin one
Thus, defining d1 and d2 by DKP eigenfunction, Ψ, can be easily calculated using the ex-
ϕ2 + ϕ1 pressions (7)-(9), whose non-relativistic limit is obtained taking
d1 = (27) E =  + m2 c4 with  << mc2 .
2
ϕ2 − ϕ1
d2 = , (28)
2i Special cases
it follows that the equations (25)-(26) can be rewritten by the
form Below we explore the special cases associated to the situa-
   tions where d1 = 0, d2 = 0, d1 = ±d2 , ω = ±e ω. The angular
E 2 − m2 c4 ϕ1 = c2 p2 + m2 (ω + ωe )2 r 2 frequency of the oscillator ω is assumed to be positive while the
qB
 magnetic field frequency ω e = 2mc can be positive or negative
−2m~ (ω + ω e) − 2m (ω + ω
e) Lz ϕ1 (29) depending on the sign of qB. For ω e > 0 (resp. < 0) it follows
3
that ω2 < ω0 < ω1 (resp. ω1 < ω0 < ω2 ) which shows that the n=0, l=1
change of sign of ω e flips the frequencies of the vectorial sector. 0.4
Cases d1 = 0 or d2 = 0 0.3
In these cases we consider the equations (25)-(26) in order to ρ0,1(1)
obtain: ϕ1 = −ϕ2 for d1 = 0 or ϕ1 = ϕ2 for d2 = 0. Hence, for 0.2

ρi
ρ0,1(0)
both cases we have 0.1
  h i ρ0,1(2)
E 2 − m2 c4 ϕ1,2 = c2 p2 + m2 ω20 r2 − 2~ω − 2e ωLz ϕ1,2 . (34) 0.0
0 1 2 3 4
These cases show us the each component ϕ1,2 of the eigenfunc- ξ
tion behaves as scalar DKPO in the √ presence of the magnetic n=1, l=1
field, with angular frequency ω0 = ω2 + ω e2 . Consequently, 0.7
this situation presents the same results as pointed out in (12)- 0.6
(13) including the non-occurrence of the Zitterbewegung fre- 0.5 ρ1,1(1)
quency. This result is in complete agreement with [34]. 0.4

ρi
Case d1 = id2 0.3 ρ1,1(0)
0.2
In this case we consider (25)-(26) in order to obtain ϕ1 = 0 0.1 ρ1,1(2)
and 0.0
  h i 0 1 2 3 4 5 6
E 2 − m2 c4 ϕ2 = c2 p2 + m2 ω22 r2 − 2~ω2 + 2ω2 Lz ϕ2 . (35)
ξ
Thus, the problem is mapped onto DKPO without magnetic n=2, l=1
field but with a reduced frequency ω2 = ω − ω e and a spin pro- 0.8
jection S z = −1. 0.6 ρ2,1(1)
Case d1 = −id2
ρi

In this case we consider (25)-(26) in order to obtain ϕ2 = 0 0.4 ρ2,1(0)

and 0.2
ρ2,1(2)
0.0
  h i
E 2 − m2 c4 ϕ1 = p2 + m2 ω21 r2 − 2~ω1 − 2ω1 Lz ϕ1 . (36) 0 2 4 6 8
Thus, the problem is mapped onto DKP oscillator without mag- ξ
netic field but with an increased frequency ω1 = ω + ω e and a
Figure 2: Probability distributions of the l = 1 eigenstates (41) for (a) n = 0
spin projection S z = +1. qB
(top), (b) n = 1 (center) and (c) n = 2 (bottom) with ω = 2e ω = 2mc > 0 and
Case ω e=ω ξ = r/a. The splitting implies the case −ω = 2e
ω < 0 is obtained by exchanging
In this case we just need make ω = ω e in the (31), thus ob- s1 and s2 (ρ1 ↔ ρ2 ), which means to reverse the direction of B.
taining
    √
E 2 − m2 c4 ϕ1 = c2 p2 + 4m2 ω2 r2 − 4~ω − 4ωLz S z ϕ1(, 37) mωa2
~ = 1 so we have ω1 = ω(1 + 21 sign(e ω)), ω0 = 25 ω, and
2
ω2 = ω(1 − 12 sign(e ω)). If we define αi = mω~i a it follows that
 
E 2 − m2 c4 ϕ2 = c2 p2 ϕ2 . (38) √
α1 = 1 + 21 sign(e ω), α0 = 5/2 and α2 = 1 − 21 sign(e ω), where
Hence, the problem is mapped onto DKP oscillator without sign(γ) denotes the sign function of γ for all γ ∈ R. Their
magnetic field but with a duplicated frequency 2ω and a spin corresponding probability distributions are obtained from the
projection S z = +1 for the component ϕ1 , while the component squared modulus of (13) and (33), expressed in the compact
ϕ2 stops the oscillation and behaves as a free particle. form by
Case ω e = −ω
In this case we just need make ω = −e ω in the (31), obtaining
 2 |l| −α ξ2 2|l|  2 
n,l (ξ) = αi ξ
ρ(i)   e i Ln αi ξ , (41)
   
E 2 − m2 c4 ϕ2 = c2 p2 + 4m2 ω2 r2 − 4~ω − 4ωLz S z ϕ2(, 39) with i = 0, 1, 2, ξ = r/a a dimensionless variable and a a char-
acteristic length (whose meaning will be clear later). The de-
 
E 2 − m2 c4 ϕ1 = c2 p2 ϕ1 , (40)
pendence in θ has disappeared due to the spherical symmetry
Hence, the problem is mapped onto DKP oscillator without of the problem. In order to study the interplay between the
magnetic field but with a duplicated frequency 2ω and a spin splitting and the energies, from (12) and (32) we can recast the
projection S z = −1 for the component ϕ2 , while the component energies for the splitting as
ϕ1 stops the oscillation and behaves as a free particle.
(i) 2
(n,l ) =
ω+ω)
Analysis of the results 1− 2~(|l|e
mc2
δ0i + ~ωi
mc2
(4n + 2|l| (1 − si ) + 2(si + δ0i )) , (42)

Now we analyze the energies and the eigenstates proba- where n,l
i
are the energies adimensionalized by mc2 , δ0i the
ω| = ω and
bility distributions. For fixing ideas we set 2|e Kronecker delta, i = 0, 1, 2 and s0 = 0, s1 = −s2 = 1. For
4
 2
compatibilizing with mωa ~ = 1 we set mc
~ω
2 = 1 so the charac-

teristic length a results to be the Compton wavelength a = mc~

,
which is a natural representation for mass on the quantum scale.
ϵn,l (1) The formula (42) turns out
100
(i) 2
(n,l ) =1
80 22.5
ω) + 2)δ0i + αi (4n + 2|l| (1 − si ) + 2(si + δ0i )) . (43)
−(|l|sign(e

17.5 In Fig. 2 we show the probability distributions (41) of the l = 1

60
eigenstates ϕi with i = 0, 1, 2 for n = 0, 1, 2 and ω e > 0. We
12.5 can see that the distributions of the spin projections s1 = 1 and
l

40
7.5
s0 = 0 are close while the corresponding to s2 = −1 exhibits a
different behavior, as a consequence of the splitting frequencies.
20 2.5 In Fig. 3 we show the energies (43) of the eigenstates ϕi for 0 ≤
n, l ≤ 100 and ω e > 0. The straight lines of Fig. 3 correspond
0 to the degenerated states whose slopes are only depending on
0 20 40 60 80 100
the sign of ωe, that is, they are in function of the sign of B. In
n fact, by letting ε(i)
n,l = ki = constant in (42) the curves l = l(n) of
ϵn,l (0) degeneracy in the range l ≥ 0 have the slopes
100 2
− , for s0 = 0
1− √ ω
e
80 ω2 +e
ω2
22.5
2
− , for s1 = −s2 = 1 (44)
17.5 1 − si
60

12.5 where ω1,2 , 0. From (42) some more features about the ener-
l

40 gies can be highlighted. For the component of projection s1 = 1

7.5
the vibrational (l = 0) and rotational (l , 0) states with quan-
20 2.5 tum number n have identical energy, so the magnetic field turns
off the angular momentum and spin effects. It is also worth not-
0 ing that in virtue of the splitting, the case ω
e < 0 has the same
0 20 40 60 80 100 eigenfunctions and energies but with the components flipped
n ϕ1 → ϕ2 , while ϕ0 remains invariant.
ϵn,l (2)
100 Thermodynamics of the vectorial sector

We illustrate the effect of the splitting in the statistical proper-

80 18 ties of the vectorial sector of the DKP oscillator since the scalar
14
sector does not exhibit a cancellation of the oscillation. For
60
accomplish this, we consider that the system is at equilibrium
10
l

with a thermal bath of finite temperature T in order to obtain the

40 partition function of its canonical ensemble. Also, we consider
6
only the states with positive energy due to those with negative
20 2 energy are unlimited below, thus ensuring a stable ensemble
[41]. Then, the partition function results
0    
0 20 40 60 80 100
X
 X

Z =  exp −γyn1 ,l1  ×  exp −γyn2 ,l2 
 (45)
n n1 ,l1 n2 ,l2
2
Figure 3: Contour plots of the splitting energy levels (43): (a) s1 = 1 (top), (b) with γ = mc
kB T and
qB
s0 = 0 (center) and (c) s2 = −1 (bottom) for ω = 2e ω = 2mc > 0 within the
range 0 ≤ n, l ≤ 100, . Due to the symmetry of (43) around l = 0, i.e. n,l (i)
= y2ni ,li = 1 + αi (4ni + 2|li | (1 − si ) + 2si )) , i = 1, 2. (46)
(i)
n,−l , only the positive range of l is shown. The lines indicate the degeneracy
Here kB is the Boltzmann constant and we have employed again
whose slopes are, from (44), ∞, − 2
and −1 for s1 = 1, s0 = 0, and s2 = −1.
1− √1
5 αi = mc
~ωi
2 with ωi = ω + si ω
e for i = 1, 2. For facilitating the
calculations we make mc2 = 1/2 and then αi = 21 (1 + si δ)
~ω

with ω
e = δω. We also consider δ ∈ [0, ∞) to avoid negative
expressions in the right hand of (46) and thus to simplify the
5
counting of degeneracies. Since y2n1 ,l1 = 2(1 + δ)n1 + 2 + δ and Thus, the partition function (53) results expressed in function
2
y2n2 ,l2 = 2(1 − δ)(n2 + |l2 |) + δ we can calculate the sums of (45) of δ and γ = kmc BT
, with the latter measuring the ratio between
separately by means of the general formula the rest mass and the thermal energy. It is worth noting that
X Z(γ, δ) , Z(γ, −δ) due to the spectrum of ϕ1 and ϕ2 is not
Ω(Ek ) exp (−βEk ) , (47) flipped when δ → −δ. In the high temperature limit T → ∞
k
we have Z(γ, δ) ≈ (1+δ)6 2 γ4 (1−δ)6 2 γ4 and then the partition function
where Ω(Ek ) indicates the degeneracy of the energy level Ek . results symmetric in δ. On the other hand, the partition function
For avoiding the infinite degeneracy of yn1 ,l1 in l1 we assume (53) is real for 0 ≤ δ ≤ 1 and when δ > 1 results complex, pre-
that Ω(En1 ,l1 ) = nn11 =−l = 2n1 , which physically means that
P
senting divergences in all the interval (1, ∞). In view of (52) we
only the states with |l1 | ≤ n1 contribute significatively to En1 ,l1 can see that the individual partition functions of ϕ1 and ϕ2 have
and the rest of terms can be neglected. Then, we have the same structure ∝ 1/((1 + si δ)2 γ4 ), thus expressing that only
X ∞ X
X n  p  the component ϕ2 experiments a phase transition when δ → 1.
= exp −γ 2(1 + δ)n + 2 + δ This corresponds to the particle free case ω e = ω with ω2 → 0,


exp −γyn1 ,l1
n1 ,l1 n=0 l=−n as pointed out previously. From the thermodynamic relations in
∞
X  p  function of the dimensionless variable γ [41]
= 2 n exp −γ 2(1 + δ)n + 2 + δ (.48)
∂ ln Z
n=0 U = −mc2 (internal energy)
For yn2 ,l2 we have a double degeneracy given by all the pairs ∂γ
(n2 , l2 ) such that n2 + l2 = k ∈ N0 with n2 , l2 = 0, . . . , k. This 1
F = −mc2 ln Z (Helmholtz free energy)
implies that Ω(En2 ,l2 ) = 2 k+1 1 = 2(k + 1) for βEn2 ,l2
P
= k. γ
Then, we have ∂F
S = k B γ2 (entropy)
∂γ
n2 ,l2 exp −γyn2 ,l2 =
P

 √ ∂U
= −kB γ2

C (specific heat), (54)
2 ∞ n=0 (n + 1) exp −γ 2(1 − δ)n + δ .
P
(49) ∂γ
The sums (48) and (49) can be calculated with the help of the we derive the thermodynamics of the vectorial sector of the
Euler-Maclaurin’s formula (employed in relativistic contexts DKP oscillator. Substituting (53) in (54) we illustrate in Fig.
for instance in [41, 42]) 4 the internal energy, the entropy and the specific heat. We can
∞ Z ∞ ∞ see that the thermodynamic potentials rapidly converge to their
X 1 X 1
f (n) = f (0) + f (x)dx − B2p f 2p−1 (0), (50) asymptotical behaviors when 1/γ → ∞(T → ∞), in virtue of
n=0
2 0 p=1
(2p)!
that the thermal excitations erase the particularities of the vec-
with B2p the Bernoulli’s numbers and f 2p−1 (0) the derivatives torial sector spectrum as soon as the only relevant term in the
of odd order of f (x) at x = 0. Now the crucial observation is total partition function (53) is proportional to T 8 . From (53) the
2
that in the limit of high thermal excitations γ = mc
kb T  1 since
asymptotical expressions of (54), that also show the additivity
the first and third terms of (50) only have powers of γ and the property in the vectorial sector, are given by
integral has powers of γ−1 , then it is enough to consider only U ≈ 8kB T = 4kB T + 4kB T = U1 + U2
the integral , i.e. !
kB T
 R∞  √   F ≈ −8kB T ln =
Z ≈ 2 0 x exp −γ 2(1 + δ)x + 2 + δ dx × mc2 (1 − δ2 )1/4
 R∞  √   ! !
2 0 (x + 1) exp −γ 2(1 − δ)x + δ dx for γ  1. (51) kB T kB T
−4kB T ln − 4k B T ln
mc2 (1 + δ)1/4 mc2 (1 − δ)1/4
Using in (51) that for a, b, γ ≥ 0
= F1 + F2
√  √ 
4e−γ b
bγ2 + 3 bγ + 3
! !
Z ∞ √ kB T
+ =
 
x exp −γ ax + b dx = S ≈ 8kB ln
mc2 (1 − δ2 )1/4
1
0 a2 γ4 ! ! ! !
Z ∞  √ kB T kB T
+ + +

(x + 1) exp −γ ax + b dx = 4kB ln 1 4k B ln 1
0 mc2 (1 + δ)1/4 mc2 (1 − δ)1/4
√ √
C ≈ 8kB = 4kB + 4kB = C1 + C2 . (55)
  
2e−γ b bγ aγ2 + 6 + aγ2 + 2bγ2 + 6
, (52)
a2 γ4
Conclusions
we obtain the partition function of the vectorial sector for γ  1
√  √  We have revisited the (2 + 1)-dimensional DKP oscillator
 2e−γ 2+δ (2 + δ)γ2 + 3 2 + δγ + 3 

Z(γ, δ) =    × in an external magnetic field from scalar 4 × 4 and vectorial
(1 + δ)2 γ4 
6 × 6 representations, which allow to study several cases of the
 √ √  literature as well as calculating their energies and eigenfunc-
 e−γ δ δγ (1 − δ)γ2 + 6 + (1 − δ)γ2 + 2δγ2 + 6 
 
  . (53) tions, in a unified way. The energies and eigenfunctions for
 (1 − δ)2 γ4 
the scalar DKPO in a uniform magnetic field are shown in the
6
 15 Acknowledgments
10 The authors acknowledge support received from the Na-
5 δ0 tional Institute of Science and Technology for Complex Sys-
U

0 tems (INCT-SC), and from the CNPq and the CAPES (Brazilian
-5 δ  0.9 agencies) at Universidade Federal da Bahia, Brazil.
-10 δ  1.1
0.0 0.5 1.0 1.5 2.0 References
1
[1] R. J. Duffin, Phys. Rev. 54, 1114 (1938).
γ [2] N. Kemmer, Proc. R. Soc. A 173, 91 (1939).
[3] G. Petiau, Acad. R. Belg. Cl. Sci. Mém. Collect. 8, 16 No 2 (1936).
30 [4] R. A. Krajcik and M. M. Nieto, Am. J. Phys. 45, 818 (1977).
[5] E. M. Corson, in Introduction to Tensors, Spinors, and Relativistic Wave-
20 Equations, (Blackie and Sons, London, 1953).
δ0
S

[6] M. Montigny and E. S. Santos, J. Math. Phys. 60, 082302 (2019).

10 [7] H. Umezawa, in Quantum Field Theory, (North-Holland, Amsterdam,
0 δ  0.9 1956).
[8] J. T. Lunardi, B. M. Pimentel, R. G. Teixeira, and J. S. Valverde, Phys.
-10 δ  1.1 Lett. A 268, 165 (2000).
0.0 0.5 1.0 1.5 2.0 [9] R. F. Guertin and T. L. Wilson, Phys. Rev. D 15, 1528 (1977).
[10] L. Kurth Kerr, B. C. Clark, S. Hama, L. Ray, and G. W. Hoffmann, Prog.
1 Theor. Phys. 103, 321 (2000).
γ [11] V. Gribov, Eur. Phys. J. C 10, 71 (1999); 10, 91 (1999).
[12] V. Ya. Fainberg and B. M. Pimentel, Theor. Math. Phys. 124, 1234 (2000).
8 [13] J. T. Lunardi, L. A. Manzoni, B. M. Pimentel, and J. S. Valverde, Int. J.
Mod. Phys. A 17, 205 (2002).
6 [14] L. M. Abreu, E. S. Santos, and J. D. M. Vianna. J. Phys. A: Math. Theor.
δ0 43, 495402 (2010).
C

4 [15] R. Casana, V. Ya. Fainberg, B. M. Pimentel and J. S. Valverde, Phys. Lett.

δ  0.9 A 316, 33 (2003).
2 [16] L. M. Abreu, A. L. Gadelha, B. M. Pimentel and E. S. Santos, Physica A
δ  1.1 419, 612 (2015).
0 [17] H. Belich, E. Passos, M. D. Montigny, and E. S. Santos, Int. J. Mod. Phys.
0.0 0.5 1.0 1.5 2.0 A 33, 1850165 (2018).
1 [18] R. Casana, C. A. M. de Melo, and B. M. Pimentel, Class. Quantum Grav.
γ 24, 723 (2007).
[19] A. Boumali. Can. J. Phys. 85, 1417 (2007).
Figure 4: (a) Internal energy (top), (b) entropy (center) and (d) specific heat [20] J. A. Swansson et al., J. Phys. A: Math. Gen. 34, 1051 (2001).
(bottom) of the vectorial sector of the two-dimensional DKP oscillator in an [21] M. Montigny, F. C. Khanna, A. E. Santana, E. S. Santos, and J. D. M.
external magnetic field B = Bẑ in function of the real parameter δ ∈ [0, ∞), Vianna, J. Phys. A: Math. Gen. 33, L273 (2000).
qB 2 [22] M. Montigny, F. C. Khanna, A. E. Santana, E. S. Santos and J. D. M.
with δ = mcω and ω = mc 2~ the oscillator frequency. The discontinuity of the
dashed curves indicate the phase transition that occurs around δ = 1 due to the Vianna, J. Phys. A: Math. Gen. 34, 8901 (2001).
cancellation of the oscillation of the 2-component. [23] E. S. Santos and L. M. Abreu, J. Phys. A: Math. Theor. 41, 075407 (2008).
[24] Y. Nedjadi and R. C. Barret, J. Phys. A: Math. Gen. 27, 4301 (1994).
[25] A. Boumali. J. Phys. A: Math. Gen. 42, 235301 (2009).
[26] M. C. B. Fernandes and J. D. M. Vianna. Braz. J. Phys. 29, 487 (1998).
equations (12)-(13), respectively,
√ where the angular frequency [27] B. Boutabia-Chéraitia and T. Boudjedaa. Phys. Lett. A 338, 97 (2005).
qB
ω is replaced by ω0 = ω + ω 2 e being ω
2 e = mc the field an- [28] P. Ghose, M. K. Samal and A. Datta, Phys. Lett. A 315, 23 (2003).
[29] J. T. Lunardi, B. M. Pimentel and R. G. Teixeira. Gen. Relativ. Gravit. 34,
gular frequency. The vector DKPO interacting with a magnetic 491 (2002).
field presents a splitting ω → (ω1 , ω0 , ω2 ) in the frequency of [30] H. Hassanabadi, B. H. Yazarloo, S. Zarrinkamar, and A. A. Rajabi, Phys.
the oscillation (Fig. 1), that corresponds to the spin projec- Rev. C 84, 064003 (2011).
tions si = +1, 0, −1 of the vector DKP field. The energies and [31] B. Mirza, R. Narimani and S. Zare, Commun. Theor. Phys. 55, 405 (2011).
[32] L. B. Castro and A. S. Castro, Phys. Lett. A 375, 2596 (2011).
the eigenfunctions of this oscillator are presented in (12), (13), [33] H. Hassanabadi, Z. Molaee and S. Zarrinkamar, Eur. Phys. J. C 72, 2217
(32), (33) and some of them illustrated in Figs. 2, 3, where (2012).
the slopes of the degeneracies (44) are in function of ω, ω e and [34] A. Boumali, L. Chatouani and H. Hassanabadi, Can. J. Phys. 91, 1 (2013).
si . Some special cases have been studied with two critical ones [35] L. B. Castro and A. S. Castro, Phys. Rev. A 90, 022101 (2014).
[36] S-r. Wu, Z-w. Long, C-y. Long, B-q. Wang and Y. Liu, Eur. Phys. J. Plus
when ω e = ±ω → B = ± 2mcω q ẑ. In these cases the component ϕ1 132 186 (2017).
or ϕ2 stops oscillating, where the phase transition ω e → ω has [37] Y. Chargui, Phys. Lett. A 382, 949 (2018).
been characterized by means of the canonical ensemble of the [38] M. Falek, M. Merad and M. Moumni J. Math. Phys. 60, 013505 (2019).
[39] B. P. Mandal and S. Verma, Phys. Lett. A 374, 1021 (2010).
vectorial sector. The thermodynamic potentials exhibit a rapid [40] V. Tyagi, S. K. Rai and B. P. Mandal, EPL 128, 30004 (2019).
convergence to their asymptotical expressions due to the ther- [41] M. H. Pacheco, R. V. Maluf, C. A. S. Almeida and R. R. Landim, EPL
mal excitations (Fig. 4), thus resulting the partition function 108, 10005 (2014).
(53) ∝ T 8 and the individual ones ∝ T 4 , with the divergences [42] K. Nouicer, J. Phys. A: Math. Gen. 39, 5125-5134 (2006).
associated to the critical values of the magnetic field.
7